Strategy vs AI

Strategy is making future predictions. Then using the results in plotting a course of action for a goal decided by you or a committee. AI does not do strategy - yet.  Why? Goals. Ultimately Goals are decided by people. AI is given goals by people, for example :

• Drive me to XYZ
• Present product ABC to customers most likely to buy it
• Create a picture based on these prompts

Notice the pattern. AI is being told what to. But not how to do it.

As AI becomes more capable these goals will be become more and more abstract. 

• Reduce churn by 10%
• Increase Customer satisfaction by 30%
• Design a new home for me

AI will devise its own path to achieve your Goal. However, You are still responsible for defining your goals which collectively represent your life strategy. 

Future Requests to AI engines may look like this :

• Prepare a plan for me to live my best life
• Run my business - alert me if there are any huge issues
• What should I do now?

So the question becomes - will AI eventually decide your strategy for you? Will it give you goals which benefit you.  Or will it give you goals which benefit someone else or itself?

The Decision Framework Advantage

 Tools plus Skills. Skills plus Tools. 

You may be the best race driver in the world. However you will never beat a below average driver with a tricycle when they have a full-blown red-bull F1 car.

You may be the best fighter pilot  in the world. However you will never beat a below average pilot using a glider and peashooter when they have a loaded F-35 aircraft.

Tools matter.

An organisations decision framework is also a tool. A tool which impacts revenue and profitability. It also impacts the lives of its employees. The lower the rank the higher the impact. A C-grade CEO with an excellent existing decision framework will be many years ahead of an A-grade CEO who needs to create one.

Paternalistic vs Predatory Marketing

Paternalistic  

Marketing for engagement based on positive values such as : 

• Relevant time saving advice
• Contributing to the community 
• Promotion of products you need

Predatory 

Marketing for engagement based on negative values such as :

• Outrage, Anger and Gossip
• Encouraging addictive traits
• Promotion of products you want

Predatory marketing currently dominates. Paternalistic Marketing is the future.

Adding Complexity to keep the job

Complexity by design is degradation by design. 

Elegance and Style are simple. Great design is to simplify.

Taking the complex and simplifying it while enhancing utility is evolution.

Complexity by design is complexity to keep power and keep a job.

"Advertisers and publishers want to use machine learning (ML) to improve addressability and audience experience by placing ads near contextually-relevant content. For data scientists, this often requires spending significant time on data engineering, data pre-processing, and training in order to extract metadata from petabytes of media content, or score billions of ad events per day with ML models..." 

How effective is this, when you the customer is still seeing Spam upon spam.

Simple and Straight forwards relevancy check

Use complex predictive models ?

Use convoluted business rules ?

No, just ask the customer - "Are we still relevant to you?"

JFrog asked me this, and I replayed "No". An awesome customer experience!

Banality of AI Complexity

Why are AI tools complex? Is it so people can keep their jobs?  Job security through complexity?

Complex tools are hard to use effectively. Complexity reduces their potential

A huge population can use a Pencil. A very small percentage can sketch like Michelangelo. 

A small population can use AI programming tools. None can use it to their full potential.

Simplicity is important. 

Fundamental Technology Assessment

Time. Time is life. It is the most important measure of everything.

This is why performance benchmarking is a critical part of technology assessment. How fast does it run to get the job done.

But there is more to it. Getting the job done means building a solution as well as running it. Design-time and run-time. Benchmarks need to measure both those aspects.

Indeed if design-time performance did not matter everyone should comfortable developing logic in assembler of binary. Usability is important.